|Publication number||US5416749 A|
|Application number||US 08/165,035|
|Publication date||May 16, 1995|
|Filing date||Dec 10, 1993|
|Priority date||Dec 10, 1993|
|Publication number||08165035, 165035, US 5416749 A, US 5416749A, US-A-5416749, US5416749 A, US5416749A|
|Inventors||Kenny K. Lai|
|Original Assignee||S3, Incorporated|
|Export Citation||BiBTeX, EndNote, RefMan|
|Patent Citations (4), Referenced by (36), Classifications (18), Legal Events (6)|
|External Links: USPTO, USPTO Assignment, Espacenet|
1. Field of the Invention
The invention relates to electronic circuits, particularly to an improved technique for retrieving data from sequential-access memory devices.
2. Description of Background Art
In the area of electronic circuit design, various memory devices are used for storing digital data. Data stored in conventional memory devices are accessible either randomly (i.e., from various storage locations in the memory device, which are not necessarily sequentially located) or sequentially (i.e., from consecutive, neighboring storage locations).
Generally, faster access times are achievable from sequential access-type devices rather than from random access-type devices, particularly when data are read from sequential storage locations. However, as system performance requirements continue to increase, there is a need to provide faster techniques for retrieving data from sequential-access memory devices.
The invention resides in a sequential-access memory device having storage cells located at consecutively addressable rows, wherein data retrieval is achieved by enabling consecutive rows during a given cycle and selecting an enabled row using a multiplexer coupled to such rows over separate data lines provided in the memory device.
Preferably, a flip-flop, which is toggled by read and clock signals, applies a select signal to the multiplexer to select a particular data line. Alternately, the multiplexer may receive the select signal from an exclusive-OR circuit, which receives read and count signals to generate the select signal. In this way, a flip flop may receive a data signal generated by the multiplexer.
FIG. 1 is a generalized block diagram of electronic storage system 10, including storage circuit 16 and selector circuit 12, 14, 18, 20, according to the present invention.
FIGS. 2 and 3 are simplified diagrams illustrating inventive aspects of electronic storage system 10 of FIG. 1.
FIG. 4 is a table showing representative read sequence signaling.
FIG. 5 is a representative timing diagram illustrating operation of electronic storage system 10.
FIGS. 6 and 7 are schematic diagrams of alternate circuits for accessing data from storage circuit 16, according to the present invention.
FIG. 8 is a representative timing diagram illustrating operation of storage circuit 16 using alternate circuit of FIG. 7.
FIG. 1 shows generally a block diagram of electronic storage system 10, which is a preferred embodiment of the present invention. System 10 includes select circuit portion (read counter 12, decoder 14, toggle flip flop 18, and multiplexer 20) and storage circuit portion (first-in/first-out (FIFO) circuit 16).
Preferably, the storage circuit portion may be any electronic storage or memory device or combined circuitry, such as a FIFO circuit or "burst-mode" memory device, having multiple digital cells or storage registers that are accessible or readable at sequential or consecutive address locations in such circuit 16.
Thus, for example, in a preferred FIFO implementation, output data signals 44, 46 representative of data stored in sequential address or row locations are generated from FIFO circuit 16 by applying thereto corresponding addressing signals 30 to read from such sequential locations.
To provide sequential addressing signals 30, read counter 12 and decoder 14 are coupled to FIFO circuit 16. In particular, read or read-next ("RDNXT") signal 22, clock signal 24, and reset signal 17 are applied, possibly from an external, synchronous processor (not shown), to read counter 12. In response to signals 22, 24, read counter 12 generates digitally cyclic, incrementing or decrementing count or read pointer ("RPTR") signal(s) 26, preferably according to a conventional "Grey Code" or similar signal generation algorithm. Decoder 14 then receives and conventionally decodes RPTR signals 26 to apply addressing signal 30 to FIFO circuit 16, preferably at sequentially incrementing or decrementing row address locations.
In accordance with the present invention, multiplexer 20 is coupled 44, 46 to FIFO circuit 16 to access data registers therein and thereby generate selectably 28 output data signal(s) (DO) 34. To cause multiplexer 20 to select particular data lines 44, 46 in FIFO circuit 16, toggle flip flop 18 receives RDNXT signal 22 and CLK signal 24 to generate select signal (SEL) 28, which is applicable to multiplexer 20.
FIGS. 2 and 3 are simplified schematic/block diagrams which illustrate electrical coupling between multiplexer 20 and data lines 44, 46 in FIFO circuit 16. In FIG. 2, FIFO circuit 16 is shown having a pair of storage banks 34, 36 coupled 44, 46 to multiplexer 20, which generates data signal 34. In particular, banks 34, 36 are assembled as being either odd or even, such that odd bank 34 includes registers accessible from "odd" data line 46 when certain odd row address read lines 11 (e.g., R1, R3, etc.) are enabled 30 by decoder 14, and even bank 36 similarly includes registers accessible from "even" data line 44 when certain even row address read lines 13 (e.g., R0, R2, etc.) are enabled 30 by decoder 14.
In FIG. 3, FIFO circuit 16 is shown in additional detail coupled to multiplexer 20 over even, odd data lines 44, 46. FIFO 16 includes multiple single data-bit storage cells 32, each cell having storage register 50 coupled to tri-statable buffer 48 located at particular row addresses. Each buffer 48 may be enabled or de-tristated effectively by addressing signal 30 provided by decoder 14 and received by FIFO circuit 16 as odd/even read line signals 11, 13 to allow digital data stored in selected registers 50 to be accessible by multiplexer through even/odd data lines 44, 46 coupled respectively to buffers 48 located at even/odd row addresses.
Thus, in accordance with the present invention, data may be accessed from FIFO circuit 16 by enabling simultaneously, during a read or access cycle, two or more consecutive rows (i.e., an odd row and a neighboring even row) in FIFO circuit 16, such that, for example, datum from register 50 located at enabled row R0 is accessible through even data line 44, and, at the same time, datum from register 50 located at enabled row R1 is accessible through odd data line 46.
In this interleaved manner, either even or odd data line 44, 46 may be selected 28 by multiplexer 20 to provide data signal 34. Data access time is improved significantly because data from "current" and corresponding "next " rows are stabilized and selectable immediately, without incurring access delays associated with read counter 12 or decoder 14. This performance improvement is achieved merely by activating RDNXT, CLK signals 22, 24, thereby causing flip flop 18 to toggle SEL signal 28 synchronously.
In an "asynchronous, " clock-less implementation, flip flop 18 may be toggled by a single asynchronous control signal, which functions equivalently to a logical-AND of RDNXT, CLK signals 22, 24. Also note that because read counter 12 generates RPTR signal 26 for sequential row addressing, preferably according to Grey code, signal contention arising from multiplexer 20 attempting to access non-sequential row addresses or the same row addresses is functionally impossible.
In FIGS. 4 and 5, representative read operation of the inventive storage system is illustrated. FIG. 4 shows read sequence signaling for activating row address lines to enable rows 11, 13 in even and odd banks, as well as status of SEL and DO signals 28, 34.
Initially, reset state 38 is defined by applying reset signal 17 to flip flop 18. In this way, low (i.e., "0") SEL signal 28 is generated, such that even data lines 44 are selected initially by multiplexer 20. Additionally, during reset state 38, read lines 13, 11 are set by decoder 14 through addressing lines 30 to read from even row address "R0" (the current row) and odd row address "R1" (the next row), thereby enabling addressed even and odd buffers 48 respectively to cause corresponding registers 50 to be accessible to multiplexer 20 through even and odd data lines 44, 46 respectively. Thus, when reset signal 17 is applied to flip flop 18, low SEL signal 28 is applied to multiplexer 20 to select even ("E") data line 44 and thereby generate "0" as data signal 34.
During a subsequent read cycle, SEL signal 28 is toggled to cause multiplexer 20 to select odd ("0") data line 46. At this time, even bank row enable 13 is shifted incrementally to a subsequent even row "R2", which serves as the next row 40 for reading, while previous next row "R1" now becomes the current row 42 for reading. Thus, because multiplexer 20 selects odd data line 46, "1" is generated as data signal 34.
As shown for subsequent read cycles, while SEL signal 28 is toggled (e.g., from even to odd to even, etc.,) even and odd read lines 13, 11 are updated incrementally such that at any given read cycle, a current row address or a next, neighboring row address is selectable quickly by multiplexer 20 to generate data signal 34 therefrom.
In FIG. 5, a timing diagram shows representative operation of system 10 through various digital signal transitions, including CLK signal 24, RDNXT signal 22, row address lines R0-R4, RPTR signal 26, SEL signal 28, and data signal 34. As shown, CLK signal 24 is a regular digital pulse, which when received (at rising CLK edge) in combination with RDNXT signal (high) 22 by read counter 12 and by toggle flip flop 18 causes RPTR signal 26 and SEL signal 28 to be generated respectively.
Thus, as shown during first instance 70 when CLK edge rises and RDNXT signal 22 is high, next (R1) and subsequent next (R2) row addresses are enabled according to RPTR signal 26 applied by decoder 14 to FIFO circuit 16, and SEL signal 28 toggles high to cause data signal 34 to be generated by multiplexer 20 by accessing register 50 from row address R0, where buffer 48 is enabled to couple such register 50 to multiplexer 20 through even data line 44.
In next instance 72, when current row is now assigned to R1, and R2 becomes the next row for reading, according to incremented RPTR signal 26, SEL signal 28 toggles low, thereby causing multiplexer 20 to access data from register 50 in row address R1, when buffer 48 in such row R1 is enabled to couple multiplexer 20 to such register 50 through odd data line 46.
In subsequent instances 74, 76, SEL signal 28 is toggled again and data from consecutive row addresses R2, R3, etc. are accessed, as described herein according to the present invention.
In an alternate embodiment of the present invention shown in FIG. 6, data lines D2, D3 additional to data lines D0, D1 are included in FIFO circuit 16 and couple multiplexer 20 to additional banks or sets of digital storage cells having register 50 connectable to corresponding data lines D0-D3 by enabling corresponding tri-statable buffers 48 coupled thereto.
In this multiple data-line implementation, similarly to the invention for data access described in the preferred embodiment, a current row address is defined in a given read cycle; but a first subsequent row address, a second subsequent row address, and a third subsequent row address are also defined, wherein each of such four row addresses are addressable consecutively, preferably according to the Grey-code technique used by read counter 12 described herein to generate RPTR signal 26 for causing decoder 14 to apply addressing signals 30 to FIFO circuit 16.
Additionally, in this implementation, multiple selection lines 28 are generated preferably by multiple flip flops coupled to RDNXT signal 22 and CLK signal 24 and arranged to generate selection signals which cause multiplexer 20 to toggle or select sequentially during each read cycle from each "next" data line (e.g., first cycle: select D0, second cycle: select D1, third cycle: select D2, etc.). Thus, four (or more) banks of storage cells are made available in a given read cycle to multiplexer 20, according to the present invention, such that access time is reduced significantly.
In another alternate embodiment of the present invention shown in FIG. 7, data access from FIFO circuit 16 is enabled by providing an exclusive-OR circuit 51, which receives RDNXT signal 22 and RPTR'signal 26 to generate and apply SEL signal 28 to multiplexer 20, which also receives data lines 44, 46 and generates data signal 34. Also in this embodiment, flip flop 55 is provided which is toggled by data signal 34 and clocked by CLK signal 24 to generate output signal (DO') 35.
FIG. 8 shows a timing diagram of representative signals for operating the alternate embodiment of FIG. 7, showing in particular various digital transitions of CLK signal 24, RDNXT signal 22, RPTR signal 26, data signal 34, output signal 35, RPTR' signal 27, and SEL signal 28. In this alternate embodiment, SEL signal 28 is generated, similarly as in other embodiments described herein, for causing multiplexer 20 to access selected odd or even data line 44, 46, such that data signal 34 is generated and synchronously provided by flip flop 55 as output signal DO'.
In particular during operation, at timing instances when RDNXT (high) signal 22 coincides with rising CLK edge, RPTR' signal 26, which may be a binary counter for toggling at each read, is received from read counter 12 for causing multiplexer 20 to select particular odd/even row address. Also, as before for other embodiments of the present invention, SEL signal 28 is shown as selecting either the current or next row address for data retrieval. In this alternate embodiment, access delay associated with selecting multiplexer 20 is avoided by synchronously generating output signal 35 upon receiving CLK signal 24.
|Cited Patent||Filing date||Publication date||Applicant||Title|
|US5253213 *||Oct 1, 1991||Oct 12, 1993||Mitsubishi Denki Kabushiki Kaisha||Semiconductor memory used for changing sequence of data|
|US5261064 *||Sep 8, 1992||Nov 9, 1993||Advanced Micro Devices, Inc.||Burst access memory|
|US5305281 *||Aug 6, 1992||Apr 19, 1994||National Semiconductor Corporation||Multiple array memory device with staggered read/write for high speed data access|
|US5311468 *||Mar 21, 1991||May 10, 1994||Texas Instruments Incorporated||Random access memory with a serial register arranged for quick access of a second bit from an arbitrary address|
|Citing Patent||Filing date||Publication date||Applicant||Title|
|US5598576 *||Mar 30, 1994||Jan 28, 1997||Sigma Designs, Incorporated||Audio output device having digital signal processor for responding to commands issued by processor by emulating designated functions according to common command interface|
|US5615401 *||Mar 30, 1994||Mar 25, 1997||Sigma Designs, Inc.||Video and audio data presentation interface|
|US5625603 *||Jun 7, 1995||Apr 29, 1997||Sgs-Thomson Microelectronics, Inc.||Integrated circuit with unequally-sized, paired memory coupled to odd number of input/output pads|
|US5790881 *||Feb 7, 1995||Aug 4, 1998||Sigma Designs, Inc.||Computer system including coprocessor devices simulating memory interfaces|
|US5797029 *||Jan 22, 1997||Aug 18, 1998||Sigma Designs, Inc.||Sound board emulation using digital signal processor using data word to determine which operation to perform and writing the result into read communication area|
|US5818468 *||Jun 4, 1996||Oct 6, 1998||Sigma Designs, Inc.||Decoding video signals at high speed using a memory buffer|
|US5821947 *||Nov 25, 1996||Oct 13, 1998||Sigma Designs, Inc.||Mixing of computer graphics and animation sequences|
|US5829049 *||Jan 21, 1997||Oct 27, 1998||Hewlett-Packard Company||Simultaneous execution of two memory reference instructions with only one address calculation|
|US5963505 *||Jun 26, 1998||Oct 5, 1999||Sgs-Thomson Microelectronics, S.A.||Sequential access memory with low consumption|
|US5995442 *||Jan 25, 1999||Nov 30, 1999||Kabushiki Kaisha Toshiba||Semiconductor memory device|
|US6084909 *||Jan 14, 1997||Jul 4, 2000||Sigma Designs, Inc.||Method of encoding a stream of motion picture data|
|US6128726 *||Jun 4, 1996||Oct 3, 2000||Sigma Designs, Inc.||Accurate high speed digital signal processor|
|US6138214 *||Dec 19, 1997||Oct 24, 2000||Siemens Aktiengesellschaft||Synchronous dynamic random access memory architecture for sequential burst mode|
|US6249481||Nov 4, 1999||Jun 19, 2001||Kabushiki Kaisha Toshiba||Semiconductor memory device|
|US6317382||Mar 21, 2001||Nov 13, 2001||Kabushiki Kaisha Toshiba||Semiconductor memory device|
|US6373785||Jul 30, 2001||Apr 16, 2002||Kabushiki Kaisha Toshiba||Semiconductor memory device|
|US6421096||Jun 27, 1995||Jul 16, 2002||Sigman Designs, Inc.||Analog video chromakey mixer|
|US6535456||Mar 11, 2002||Mar 18, 2003||Kabushiki Kaisha Toshiba||Semiconductor memory device|
|US6654314||Feb 6, 2003||Nov 25, 2003||Kabushiki Kaisha Toshiba||Semiconductor memory device|
|US7016988 *||Nov 4, 2003||Mar 21, 2006||Stmicroelectronics, S.A.||Output buffer register, electronic circuit and method for delivering signals using same|
|US7061827||Oct 21, 2003||Jun 13, 2006||Kabushiki Kaisha Toshiba||Semiconductor memory device|
|US7158444||Dec 13, 2005||Jan 2, 2007||Kabushiki Kaisha Toshiba||Semiconductor memory device|
|US7818504||Sep 26, 2007||Oct 19, 2010||International Business Machines Corporation||Storage system that prioritizes storage requests|
|US8711639 *||Nov 2, 2010||Apr 29, 2014||Micron Technology, Inc.||Data paths using a first signal to capture data and a second signal to output data and methods for providing data|
|US8937839||Apr 23, 2014||Jan 20, 2015||Micron Technology, Inc.||Data paths using a first signal to capture data and a second signal to output data and methods for providing data|
|US9183900||Dec 22, 2014||Nov 10, 2015||Micron Technology, Inc.||Data paths using a first signal to capture data and a second signal to output data and methods for providing data|
|US9478270||Oct 20, 2015||Oct 25, 2016||Micron Technology, Inc.|
|US20030154363 *||Feb 11, 2002||Aug 14, 2003||Soltis Donald C.||Stacked register aliasing in data hazard detection to reduce circuit|
|US20040160845 *||Oct 30, 2003||Aug 19, 2004||Stmicoelectronics Pvt. Ltd.||Method and apparatus to reduce access time in synchronous FIFOS with zero latency overhead|
|US20040174752 *||Nov 4, 2003||Sep 9, 2004||Stmicroelectronics S.A.||Output buffer register, electronic circuit and method for delivering signals using same|
|US20060152979 *||Dec 13, 2005||Jul 13, 2006||Kabushiki Kaisha Toshiba||Semiconductor memory device|
|US20090083490 *||Sep 26, 2007||Mar 26, 2009||International Business Machines Coproration||System to Improve Data Store Throughput for a Shared-Cache of a Multiprocessor Structure and Associated Methods|
|US20090083491 *||Sep 26, 2007||Mar 26, 2009||International Business Machines Corporation||Storage System and Associated Methods|
|US20120110368 *||Nov 2, 2010||May 3, 2012||Micron Technology, Inc.|
|US20130290646 *||Apr 30, 2012||Oct 31, 2013||Freescale Semiconductor, Inc.||Fifo buffer system providing same clock cycle response to pop commands|
|EP1416373A2 *||Oct 28, 2003||May 6, 2004||STMicroelectronics Pvt. Ltd||Method and apparatus to reduce access time in synchronous fifos with zero latency overhead|
|U.S. Classification||365/240, 365/219, 365/236, 365/233.18, 365/233.1, 365/230.04, 365/230.06, 365/239, 365/230.02, 365/233.17, 365/230.08|
|International Classification||G11C8/04, G06F5/10|
|Cooperative Classification||G11C8/04, G06F5/10, G06F2205/102|
|European Classification||G06F5/10, G11C8/04|
|Feb 7, 1994||AS||Assignment|
Owner name: S3 INCORPORATED, CALIFORNIA
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:LAI, KENNY KE-CHUNG;REEL/FRAME:007240/0832
Effective date: 19940124
|Aug 31, 1998||FPAY||Fee payment|
Year of fee payment: 4
|May 7, 2002||AS||Assignment|
Owner name: S3 GRAPHICS CO., LTD., CAYMAN ISLANDS
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:SONICBLUE INCORPORATED;REEL/FRAME:012852/0016
Effective date: 20010103
|Nov 15, 2002||FPAY||Fee payment|
Year of fee payment: 8
|Nov 16, 2006||FPAY||Fee payment|
Year of fee payment: 12
|Aug 25, 2007||AS||Assignment|
Owner name: SONICBLUE INCORPORATED, CALIFORNIA
Free format text: CHANGE OF NAME;ASSIGNOR:S3 INCORPORATED;REEL/FRAME:019744/0134
Effective date: 20001109